The following list of other potential sources of bias in a clinical study may aid detection of further problems.
The conduct of the study is affected by interim results (e.g. recruiting additional participants from a subgroup showing more benefit).
There is deviation from the study protocol in a way that does not reflect clinical practice (e.g. post hoc stepping-up of doses to exaggerated levels).
There is pre-randomization administration of an intervention that could enhance or diminish the effect of a subsequent, randomized, intervention.
Inappropriate administration of an intervention (or co-intervention).
Contamination (e.g. participants pooling drugs).
Occurrence of ‘null bias’ due to interventions being insufficiently well delivered or overly wide inclusion criteria for participants (Woods 1995).
An insensitive instrument is used to measure outcomes (which can lead to under-estimation of both beneficial and harmful effects).
Selective reporting of subgroups.
Fraud.
Inappropriate influence of funders (or, more generally, of people with a vested interest in the results) is often regarded as an important risk of bias. For example, in one empirical study, more than half of the protocols for industry-initiated trials stated that the sponsor either owns the data or needs to approve the manuscript, or both; none of these constraints were stated in any of the trial publications (Gøtzsche 2006). It is important that information about vested interests is collected and presented when relevant. However, review authors should provide this information in the ‘Characteristics of included studies’ table (see Section 11.2.2). The ‘Risk of bias’ table should be used to assess specific aspects of methodology that might be been influenced by vested interests and which may lead directly to a risk of bias. Note that some decisions that may be influenced by those with a vested interest, such as choice of a particularly low dose of a comparator drug, should be addressed as a source of heterogeneity rather than through the ‘Risk of bias’ tool, since they do not impact directly on the internal validity of the findings.